Vol. 60, No. 6

INFECTION AND IMMUNITY, June 1992, p. 2506-2513

0019-9567/92/062506-08$02.00/0 Copyright X 1992, American Society for Microbiology

Binding and Neutralization of Endotoxin by Limulus Antilipopolysaccharide Factor H. SHAW WARREN,l,2,3* MAUREEN L. GLENNON, 3t NORMAN WAINWRIGHT,4 STEPHEN F. AMATO,1,3 KERRY M. BLACK,"13 STEPHEN J. KIRSCH,13t GILLES R. RIVEAU,5§ RICHARD I. WHYTE,611 WARREN M. ZAPOL,7 AND THOMAS J. NOVITSKY4 1 Departments of Pediatrics, Medicine,2 Surgery6 and Anesthesia,7 Massachusetts General Hospital and Harvard Medical School, Boston, Massachusetts 02129; Department of Pediatrics, Shriners Burns Institute, Boston, Massachusetts 021143; Department of Pharmacology, University of South Flonida, Tampa, Florida 336205; and Associates of Cape Cod, Falmouth, Massachusetts 025404 Received 7 October 1991/Accepted 2 April 1992

In order to examine the ability of Limulus antilipopolysaccharide factor (LALF) to bind lipopolysaccharide (LPS), we purified LALF to homogeneity from Limulus amoebocyte lysate and coupled it covalently to agarose beads. LALF-coupled beads captured more tritiated LPS from rough and smooth strains of gram-negative bacteria than did control human serum albumin-coupled beads. Unlabeled homologous and heterologous LPS competed for the binding of 3H-LPS to LALF-coupled beads. LALF bound LPS in a dose-dependent manner as assessed by the precipitation of LPS-LALF complexes with 50% saturated ammonium sulfate. We also studied the ability of LALF to neutralize LPS. LPS preincubated with LALF was less mitogenic for murine splenocytes, was less pyrogenic in the rabbit fever assay, was less lethal in mice which had been sensitized to LPS with actinomycin D, and induced less fever, neutropenia, and pulmonary hypertension when infused into sheep. Our findings extend prior studies which suggested that LALF binds to and neutralizes LPS from multiple strains of gram-negative bacteria.

Despite the availability of antibiotics capable of rapidly killing most gram-negative bacteria, infections caused by these organisms, with the subsequent development of sepsis, shock, and multisystem organ failure, continue to be a major clinical problem. Bacterial lipopolysaccharide (LPS) is felt to play an important role in the pathophysiology of severe gram-negative infections. Accordingly, a strategy for the treatment of gram-negative sepsis has been to attempt to develop a means to neutralize or clear LPS from the bloodstream before the induction of irreversible pathology. One such approach has been to passively infuse immunoglobulin directed at LPS. Polyclonal (15, 17, 41) and monoclonal (7, 10, 18, 21) antibodies directed at the 0 polysaccharide of LPS protect against challenge with LPS in animal models. However, immunoglobulin directed against 0 polysaccharide is only protective against homologous strains of bacteria, making it difficult to utilize clinically. Polyclonal (4, 6, 17, 20, 48, 50) and monoclonal (14, 49) antibodies directed at the common core glycolipid structure on LPS have also been reported to give protection in animal models and clinical trials. These immunoglobulin preparations are reported to protect against most gram-negative strains, although the protection in animal models is less than that of antibody directed at the 0 polysaccharide. There are relatively few proteins which have been reported to bind and neutralize LPS. Two such proteins are *

Corresponding author.

t Present address: University of Vermont Medical School, Burl-

ington, VT 05405. t Present address: Downstate Medical School, Syracuse, NY 13201. § Present address: Laboratory of Experimental Immunopharmacology, Institut Biomedicale des Cordeliers, Paris, France. 1 Present address: Department of Thoracic Surgery, University of Michigan, Ann Arbor, MI 48109-0624.

polymyxin B and bactericidal/permeability-increasing protein (19, 45). Polymyxin B has been known for over two decades to protect animals against challenge with LPS (8, 31, 34-36) and against gram-negative infection (3, 9), but it is believed to be too toxic for routine clinical use. Bactericidal/ permeability-increasing protein was recently reported to inhibit LPS-induced stimulation of neutrophils and coagulation of Limulus amoebocyte lysate (LAL) (19). Further studies will be needed to assess whether it will be helpful as a therapeutic agent for endotoxemia. In 1982, an anticoagulant which inhibited the endotoxinmediated activation of the Limulus coagulation system was identified in the amoebocytes of the hemolymph of the horseshoe crabs Tachypleus tridentatus and Limulus polyphemus (40). This factor has since been isolated and characterized (22), and it is a single-chain polypeptide with a molecular weight of 11,800 (26). The primary structure of the molecule consists of 102 amino acids and is partially homologous with structures of molecules in the lactalbumin-lysozyme family (1). Several experiments suggest that this factor acts by binding to LPS. First, the factor inhibits activation of LAL by LPS, but not by another activator, (1-3)-p-D-glucan (40). Second, the factor lyses erythrocytes that are sensitized by prior incubation with LPS but does not lyse unsensitized cells, and the ability to hemolyze is inhibited by excess LPS added to the reaction mixture (29). Third, the activation of cultured human endothelial cells by LPS is decreased in a dose-dependent manner if the LPS is preincubated with the factor (11). Fourth, the factor forms precipitation lines when reacted with LPS in agarose double-diffusion gels (27). Because of these studies, the factor has been called Limulus anti-LPS factor (LALF) (1, 22). A preliminary study suggested that the lethal toxicity of LPS in rats was diminished by preincubation with LALF (43). Recently, LALF has also been shown to attenuate the 2506

NEUTRALIZATION OF LPS BY LALF

VOL. 60, 1992

toxic effects of meningococcal lipooligosaccharide in rabbits (2). To test the hypothesis that this factor has antiendotoxic properties in other systems, we purified LALF to homogeneity from LAL and examined its ability to bind to endotoxin from several different gram-negative bacteria by measuring the amount of 3H-LPS that the factor captured when covalently coupled to agarose beads and by precipitating 3HLPS-LALF complexes with ammonium sulfate. We then assessed the ability of LALF to decrease the mitogenic activity of LPS on murine splenocytes in vitro and to decrease LPS-induced pyrogenicity in rabbits, lethality in mice treated simultaneously with actinomycin D, and fever, leukopenia, and pulmonary artery vasoconstriction and hypertension in awake sheep. Our results confirm and extend prior studies and indicate that LALF binds to and neutralizes LPS from numerous different gram-negative strains both in vitro and in vivo.

MATERIALS AND METHODS LPSs. Unlabeled LPS from Salmonella typhimurium, Klebsiella pneumoniae, and Serratia marcescens were purchased from List Co. (Campbell, Calif.). Unlabeled LPS from Escherichia coli 0113 was prepared by the hot phenol method as described by Rudbach et al. (39). Cultures of S. typhimurium G30, E. coli 0111:B4, E. coli 018, and E. coli J5 were the kind gifts of Paul Rick (Uniformed Health Services University and Health Sciences, Bethesda, Md.), David Morrison (University of Kansas Medical Center, Kansas City), George Siber (Dana Farber Cancer Institute, Boston, Mass.), and Jerald Sadoff (Walter Reed Army Institute of Research, Washington, D.C.), respectively. Cultures of E. coli 04, 06, 016, 025, and 075 were the kind gifts of Alan Cross (Walter Reed Army Institute of Research). Biosynthetically radiolabeled isolates of all of the "smooth" E. coli strains were prepared by growing the organisms in the presence of 3H-acetate and then subjecting them to hot phenol extraction (44). Briefly, we grew cultures of each organism to an optical density of 0.9 at 540 nm (with a path length of 1.0 cm) in broth containing (per liter) 22.5 g of yeast extract, 11 g of peptone, 4 g of monobasic potassium phosphate, 16.8 g of dibasic potassium phosphate, and 10 g of glucose, with 10 mCi of 3H-acetate per 100 ml of broth. The cells were chilled and washed three times in saline, and the LPS was extracted by the hot phenol method (46). The preparations were then treated first with DNase and RNase and then with pronase (Sigma Chemical Co, St. Louis, Mo.) according to the method of Romeo et al. (38). The concentrations of LPS were estimated by a spectrophotometric LAL gelatin assay utilizing an E. coli 0113 LPS standard containing 10 endotoxin units per ng (lot 20; Associates of Cape Cod, Falmouth, Mass.). These results were similar to those obtained by weight. Solutions of 3H-LPS were adjusted to 1 p,g of LPS per ml (as Limulus biological activity), and counts per minute per microgram were calculated by counting a 0.4-ml volume combined with 4.5 ml of Optiflor scintillation fluid (Packard, Downers Grove, Ill.). The different LPSs contained the following counts per minute per microgram: E. coli 04, 10,490; E. coli 06, 7,200; E. coli 018, 6,150; E. coli 016, 17,100; E. coli 025, 11,040; E. coli 075, 4,700; E. coli 0111:B4, 4,040. More than 99% of each radiolabeled LPS was demonstrated to remain in the water phase after a 1:1 ether-water extraction at pH 5. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDSPAGE) of each LPS resulted in a regularly spaced band

2507

0.60 F 0.50 0

0.40

o (N

0.30

ri

0.20

C.1o

0.00

ith 0

10

20

30

40

50

60

ME (Ml N) FIG. 1. Purity of LALF as assessed by high-pertormance liquid chromatography and SDS-PAGE. Conditions for SDS-PAGE are given in Materials and Methods. The apparent molecular weight of the protein band shown was estimated to be 15,000 by comparing it with molecular weight markers (not shown). O.D., optical density.

pattern typical of LPS when stained with silver. Similar regularly spaced band patterns were obtained when the gels were analyzed by autoradiography. Radiolabeled LPS from E. coli J5 was made by growing the organisms in broth containing 3H-acetate as described above and extracting the LPS as described by Galanos et al. (13). This LPS contained 22,000 cpm/p,g. Radiolabeled LPS from S. typhimurium G30 was prepared by growing the organisms in PPBE broth (per liter: 10 g of peptone, 1 g of beef extract, and 5 g of NaCl in 0.05 mM unlabeled D-galactose) containing 500 1Ci of D-[1-3H]galactose per 100 ml of broth as described by Munford and Hall (23). This strain produces complete LPS only in the presence of galactose and incorporates exogenous galactose almost entirely into the LPS (30), thus ensuring that all of the extracted label is found in LPS. The LPS was then extracted and standardized as described above. This LPS contained 16,800 cpm/,ug. Purification of LALF from L. polyphemus amoebocytes. LALF was purified from LAL as previously described (43) by using a spectrophotometric LAL assay to monitor inhibition of LPS-induced lysate activation (28). Briefly, amoebocytes from L. polyphemus were collected under endotoxin-free conditions, lysed by the addition of distilled water, and centrifuged at 5,000 x g for 30 min. The pellet was extracted with 3 M urea. The extract was filtered through a membrane with a 30,000-Da cutoff and concentrated by a membrane with a 8,000-Da cutoff. The retentate was applied to a cation exchange column (CM Sepharose) equilibrated with 3 M urea-10 mM ammonium acetate (pH 5.5) and step eluted with NaCl at 0.15, 0.25, 0.5 M. The 0.5 M NaCl peak was directly applied to a C-4 reversed-phase column (Vydac, Hesperia, Calif.) equilibrated with water-0.2% trifluoroacetic acid. The column was step eluted with 25, 35, and 50% isopropanol containing 0.2% trifluoroacetic acid. The 50% isopropanol peak was lyophilized and reconstituted immediately before use. The final product was estimated to be >95% pure by reversed-phase high-performance liquid chromatography and SDS-PAGE (Fig. 1). SDS-PAGE was carried out with a Pharmacia PhastSystem at 8 to 25% acrylamide with 2% cross-linking and the manufacturer's Tris acetate buffer system. Staining was performed with Coomassie

2508

WARREN ET AL.

R 350. The amino acid sequence of LALF was determined up essentially identical to that

to residue 30 and was found to be reported by Muta et al. (26).

Binding assays. The ability of LALF to bind LPS was assessed by two methods. In the first method, we measured the amount of LPS that was captured by agarose beads to which LALF had been covalently coupled. Control beads were coupled with human serum albumin (HSA). LALF and HSA were covalently coupled to carbonyldiimidazole-activated agarose gel beads (Reacti-gel; Pierce, Rockford, Ill.) according to the manufacturer's directions. Briefly, the resin was washed and equilibrated in 0.1 M borate-0.9% NaCl, pH 8.5. Ten milligrams of protein was added per ml of packed bead volume, and the mixture was mixed end-over-end for 20 h at 4°C. The coupled gel was blocked with 2 M Tris buffer (pH 8.0) for 4 h and washed with pyrogen-free distilled water. By measuring protein concentrations after coupling, 8.5 mg of LALF and 9.2 mg of HSA were coupled per ml of gel. Binding assays were performed by combining triplicate samples of 1.0-pLg/ml radiolabeled LPS with a 2% (vol/vol) solution of LALF- or HSA-linked beads in a total of 0.4 ml of phosphate-buffered saline (PBS) (pH 7.0) in 1.5-ml polypropylene conical tubes. After 120 min of end-over-end mixing at 37°C, the tubes were centrifuged for 15 min at 10,000 x g. The supernatants were manually aspirated and saved and the pellets were washed twice with PBS. Counts per minute in the supernatants and pellets were measured in a liquid scintillation counter, adjusting for quenching by the internal standard method. The percentage of LPS bound to the beads was determined by dividing the total counts per minute in the pellet by the total counts per minute recovered and multiplying by 100. The total counts per minute recovered varied slightly depending on the type and quantity of LPS added and whether LALFor HSA-coupled beads were used. Typically, 60 to 80% of added counts were recovered. Counts per minute not recovered with the beads or supernatant were quantitatively accounted for and shown to be bound to the tube walls by cutting the polypropylene tubes into quarters with a razor and counting in a similar manner. Competition binding experiments were performed by adding unlabeled homologous or heterologous LPS to reaction mixtures containing LALF-coupled beads and 1.0 p,g of 3H-LPS per ml. Results are expressed as means ± standard deviations. In the second method, 3H-LPS was incubated in dilutions of LALF at 37°C. Complexes of 3H-LPS bound to LALF were then separated from free 3H-LPS by precipitation in half-saturated ammonium sulfate according to the method of Farr (12). Specifically, 6,000 cpm of each smooth LPS tested was incubated in 150 pl of twofold dilutions of LALF in 0.02 M PBS for 120 min at 37°C in a 1.5-ml Eppendorf microcentrifuge tube. This corresponds to the following LPS concentrations in the final protein mixture: E. coli 04, 3.8 ,g/ml; E. coli 06, 5.6 p1g/ml; E. coli 016, 2.3 p,g/ml; E. coli 018, 6.5 p,g/ml; E. coli 025, 3.6 p,g/ml; E. coli 075, 8.5 ,g/ml. We elected to fix the number of counts per minute added rather than the concentration of LPS added in order to study the minimal concentration of each LPS that could be reliably detected. Following incubation at 37°C, the solution was cooled on ice for 15 min, an equal volume of iced saturated ammonium sulfate was added dropwise, and the solution was allowed to sit at 4°C for another 15 min. The tubes were then centrifuged at 12,000 x g for 15 min. Supernatants were carefully aspirated. Pellets were washed twice with 50% ammonium sulfate and resuspended in 300 ,ul of PBS. The

INFECT. IMMUN.

quantity of LPS in the supernatants and pellets was assessed by counting 0.4 ml of a 1:7 dilution of each combined with 4.5 ml of Optiflor scintillation fluid (Packard). Quenching was minimal and was corrected for by the internal standard method. Less than 5% of each smooth LPS tested precipitated in PBS alone. However, more than 90% of the 3H-LPS from rough mutant E. coli J5 precipitated in PBS-HSA alone, so that we were unable to test this LPS in the system. For most of the assays, recovery of added counts per minute was greater than 85% by this assay. At low protein concentrations of LALF or HSA (less than 50 ,ug/ml), recovery was sometimes less, which we attributed to LPS binding to the walls of the tube. All assays were performed in duplicate, and the results are given as means. Results are expressed as the percentage of recovered counts per minute as calculated by the following formula: (counts per minute in the pellet/ counts per minute recovered) x 100. Mitogenic assay. The mitogenic assay was performed essentially as described by Jacobs and Morrison (16). Briefly, dilutions of LPS were preincubated in the presence of 10 ,ug of LALF per ml for 2 h at 37°C. These mixtures were then incubated for 48 h in RPMI medium containing 5 x 106 spleen cells from CD-1 mice. One microcurie of 3H-thymidine was next added to each well, and cells were incubated for an additional 16 h. Incorporated radioactivity was measured by using a mash harvester to disrupt the cells followed by scintillation counting. Each assay was done in quadruplicate, and the results are reported as means + standard deviations. Staining of splenocytes with crystal violet following culture in the presence of 10 ,ug of LALF per ml suggested that this concentration of LALF did not kill splenocytes. Concanavalin A was normally mitogenic in the presence of 10 ,ug of LALF per ml. Rabbit pyrogen assay. The rabbit pyrogen assay has been previously described (37). Male New Zealand White rabbits (2.5 to 3.5 kg) were used throughout the study. Glassware, needles, syringes, and PBS were pyrogen free. Rectal temperature was recorded every 3 min for 5 h after intravenous injection by using Thermistor probes connected to a telethermometer interfaced to an HP85 computer (Hewlett-Packard Co., Palo Alto, Calif.). Changes in temperature are expressed as maximum deviation from the base line recorded at the time of injection. Earlier studies indicated that a dose of the same lot of E. coli 0113 LPS at 150 ng/kg of body weight was on a sensitive portion of the dose-response curve (44). Accordingly, we preincubated LPS from E. coli 0113 with a 100-fold excess of LALF by weight or in saline alone for 2 h at 37°C and administered these mixtures at an LPS dose of 150 ng/kg to groups of four acclimatized rabbits while recording their rectal temperature. Actinomycin D mouse model. Mice are relatively resistant to the lethal effects of LPS injection. We therefore utilized actinomycin D as described by Brown and Morrison (5) and Pieroni et al. (32) to sensitize mice to submicrogram amounts of LPS. Dilutions of LPS were preincubated in saline alone or in 2.0 jg of LALF per ml in saline at 37°C for 60 min and then combined 1:1 with 250 ,ug of actinomycin D per ml immediately prior to injection. A total volume of 0.2 ml of this solution containing actinomycin D (25 ,ug), LPS, and either saline or 200 ng of LALF was injected intraperitoneally into groups of six to nine CD-1 mice. Results are expressed as survivors per total mice at 72 h. The 50% lethal dose was calculated by the method of Reed and Muench (33). Statistics were calculated by logistic regression using

VOL. 60, 1992

NEUTRALIZATION OF LPS BY LALF

TABLE 1. Percentage of radiolabeled LPS captured by HSA or LALF coupled to agarose beads

70 60

% of LPS captured by: HSA-

3H-LPS

coupled beads

LALFcoupled beads

7C

26.0 13.7 31.6 59.5

typhimurium coli O111:B4 coli 0118 coli J5

± ±

± ±

1.6 1.2 1.0 10.1

60.6 49.3 89.9 97.3

± ± ± ±

2.3 1.5 1.2 0.2

50

:3 0

v)

S. E. E. E.

2509

J-

40

30

T

.

20 10

the SAS statistical system (PC SAS version 6; SAS Institute, Cary, N.C.). LPS-induced fever, neutropenia, and pulmonary hypertension in sheep. This model has been previously described (47) and represents an established model for measuring LPS effects upon pulmonary hemodynamics. The injection of nanogram-per-kilogram quantities of LPS induces reliable changes of core temperature, transient neutropenia, and mediator release with vasoconstriction and pulmonary artery hypertension. For this assay, LPS from S. marcescens was utilized because detailed dose response information was known from prior experiments (47). Briefly, LPS was preincubated with a 200-fold excess of LALF or with pyrogen-free saline for 2 h at 37°C. Immediately following the incubation, 20 ng of LPS per kg (4 ,ug of LALF per kg) was intravenously injected into awake sheep. A sterile Swan Ganz catheter had been placed in the pulmonary artery via an introducer sheath in the external jugular vein. This catheter allowed us to measure and record pulmonary artery pressure and pulmonary artery core temperature continuously. We intermittently measured pulmonary artery occlusion pressure and injected 5 ml of sterile 0°C saline to measure cardiac output. Pulmonary vascular resistance was calculated by the standard formula (pulmonary pulmonary artery occlusion pressure)/ artery pressure cardiac output. Samples were obtained at intervals for quantitation of leukocytes by using a Coulter Counter. All data for the sheep experiments were stored on a DEC LSI-11/73 computer and transferred to an IBM PC-286 computer for statistical analysis using the SAS statistical system. Core temperature, pulmonary hemodynamic, and hematologic data were statistically compared over time as well as between treatment groups by using analysis of variance for repeated measures. When, within a treatment group, effects were transient and occurred at different times, data were averaged over time and compared with those of the control population by t tests. For single comparisons, P < 0.05 was considered to be significant. All data are presented as means + standard errors. -

RESULTS HSA-coupled beads to capture radiolabeled LPS. LALF-coupled beads bound two to four times as much tritiated LPS from the smooth gram-negative strains studied than did the control HSA-coupled beads (Table 1). The LALF-coupled beads bound more than 95% of tritiated LPS from rough mutant E. coli J5, although nonspecific binding to the HSA-coupled beads was higher for this strain than for LPS from the smooth strains. Competition experiments were performed by adding unlabeled LPS to the reaction mixture containing LALF-coupled beads and 3HLPS (1.0 ,ug/ml) from S. typhimurium. The addition of both

Ability of LALF-

or

0

0.5

1

10

100

Unlabeled LPS (,ug/ml)

FIG. 2. Competition for binding of tritiated LPS to LALFcoupled beads by homologous and heterologous unlabeled LPS. Tritiated LPS (1 ,ug/ml) from S. typhimurium was added to the reaction mixture together with various concentrations of unlabeled LPS from S. typhimurium (0) or K pneumonia (A). The open diamond to the right of the figure represents the percentage of 3H-LPS bound to LALF-coupled beads in the presence of 1.2 mg of albumin per ml.

homologous and heterologous unlabeled LPS decreased the percentage of radiolabeled LPS bound to the beads (Fig. 2). Ability of LALF to bind LPS as assessed by precipitation with saturated ammonium sulfate. LALF bound to 3H-LPS

from six smooth E. coli strains in a dose-dependent manner (Fig. 3). Fifty percent of the recovered LPS was bound at LALF concentrations ranging from 250 to 700 jig/ml. When corrected for LPS concentration, this corresponds to the following ratios of LALF to LPS needed for half-maximal precipitation of each LPS: E. coli 04, 137:1; E. coli 06, 118:1; E. coli 016, 226:1; E. coli 018, 52:1; E. coli 025, 72:1; E. coli 075, 65:1. Effect of LALF on mitogenic activity of LPS. E. coli 0113 LPS which had been preincubated with LALF was less mitogenic than the saline control (Fig. 4). Similar findings were obtained when S. typhimurium LPS was used (data not shown). Effect of LALF on LPS-induced fever in rabbits. E. coli 0113 LPS which had been preincubated with LALF induced significantly less fever in rabbits than did LPS incubated in saline (P < 0.001) (Fig. 5). Protective effect of LALF on actinomycin D-sensitized mice. The effect of LALF on mice sensitized with actinomycin D is shown in Table 2. As noted in Materials and Methods, mice received 25 p,g of actinomycin D, dilutions of LPS, and either saline or 200 ng of LALF. Three experiments were performed, each with a range of LPS dilutions. In each experiment, more mice survived at critical LPS doses. The calculated 50% lethal dose obtained by using all mice was 35.0 ng per mouse for LPS in the presence of LALF and 2.8 ng per mouse in the presence of saline alone. By logistic regression with additive effects on the logit scale for log LPS dose and administration of LALF, both effects were highly significant (P < 0.001). The odds ratio for the probability of survival with the administration of LALF was 1.82. Effect of LALF on LPS-induced fever, neutropenia, and pulmonary hypertension in sheep. Preincubation with LALF resulted in significantly less LPS-induced fever, neutropenia, pulmonary vasoconstriction, and hypertension than was

2510

INFECT. IMMUN.

WARREN ET AL.

100 90 0

o

0-

80

Qa

70

b0

60

a.

50

'a a. -J

I

40

C

30

*

20 0

20

10 0

103

10'

101

104

LALF or HSA Concentration (ug/ml) FIG. 3. Percentage of 3H-LPS precipitated by different concentrations of LALF (solid figures, solid lines) or HSA (open figures, dashed lines) in 50% saturated ammonium sulfate as described in Materials and Methods. Symbols and denoted strains of E. coli are as follows: 04, plus signs; 06, circles; 016, triangles; 018, squares; 025, diamonds; 075, inverted triangles.

seen with control sheep (Fig. 6). The febrile response to LPS which had been preincubated with LALF was monophasic rather than biphasic, and the peak fever was lower and delayed (P < 0.001). Similarly, the increase in pulmonary artery pressure induced by LPS was markedly diminished and delayed by preincubation with LALF (P < 0.05). Preincubation with LALF abolished LPS-induced neutropenia (P < 0.001).

DISCUSSION The major finding of our experiments is that LALF binds to LPS derived from several different gram-negative bacterial strains in vitro and that it can neutralize the toxic effects of LPS in in vitro and in vivo assays of biological activity. Our results extend prior studies using LALF in several ways. First, we have confirmed by direct binding assays that LALF binds to LPS from multiple strains of gram-negative bacteria. Second, we have shown that preincubation with

LALF diminishes the mitogenic response of LPS for murine splenocytes. Third, we have shown that LALF decreases the bioactivity of LPS in mice (which have been sensitized with actinomycin D), rabbits, and sheep. Each of these animal models represents an established assay for in vivo endotoxin activity. There are a limited number of well-defined proteins which bind LPS and are capable of these antitoxic properties. Other substances reported to bind and neutralize the effects of LPS in biological assays and/or animal models include polymyxin B (3, 8, 9, 31, 34-36), bactericidal/permeabilityincreasing protein (19, 45), and immunoglobulin directed at the endotoxin core (4, 6, 14, 17, 20, 48-50). LPS also binds to plasma lipoproteins in the presence of disaggregating agents, and the complexes which are formed are less active than unbound LPS in numerous assays (24, 25, 42). The results of the mitogenic assays, the mouse lethality test, and studies using LAL (data not shown) suggest that

o

15

1.2, 1.0

-

0

12 DA D.6

~

0.2 0 -

//

/- _~o -~--~--

0

Z;

z

0~~~~~~~~~

LU

0

1

2

3

4

5

HOURS 0.01

0.1

1

1

E3coli 01 13 LPS (ug/ml) FIG.

4.

Mitogenic response

preincubated

with LALF

of E.

coli 0113

(10 p.g/ml) (@) or

LPS which had been

saline alone

(0).

FIG. 5. Pyrogenic response in groups of four rabbits to intravenous injection of E. coli 0113 LPS (150 ng/kg) which had been preincubated with a 100-fold excess of LALF ( ) or saline alone (--- -). The maximal temperature of rabbits which were injected with LPS preincubated with LALF was significantly less than that of controls (P < 0.001).

VOL. 60, 1992

NEUTRALIZATION OF LPS BY LALF

2511

TABLE 2. Protective effect of LALF in mice sensitized with antinomycin D No. of surviving mice/total mice Dose

Expt 1

LPS/mouse (ng) 1,000.0 100.0 10.0 1.0 0o1 Saline alone

Expt 2

Totalr

Expt 3

Saline

LALF

Saline

LALF

Saline

LALF

Saline

LALF

0/6 3/6 4/6 4/6 6/6 6/6

1/6 0/6 5/6 6/6 5/6 6/6

0/6 1/6 1/6 2/6 2/6 6/6

3/6 2/6 5/6 6/6 5/6 6/6

0/9 0/9 1/9 8/9 9/9 9/9

0/9 0/9 9/9 9/9 9/9 9/9

0/21 4/21 6/21 14/21 17/21 21/21

4/21 2/21 19/21 21/21 19/21 21/21

a Fifty-percent lethal doses were 2.8 ng of LPS with saline and 35.0 ng of LPS with LALF.

approximately a 10-fold excess of LALF by weight is needed for neutralization, a relationship which has been previously shown for the inhibition of endothelial cell activation (11). Estimation of the molar ratios needed for effective neutralization is not possible owing to uncertainty as to the effective molecular weight of the micellar form of LPS in aqueous solution. + Direct comparison with polymyxin B is difficult owing to the use of different model systems. Polymyxin B has been reported to neutralize LPS when premixed in a 115:1 ratio by weight in a chick embryo lethality assay (36), in a 300:1 ratio

in a murine lethality assay in which the mice were sensitized by adrenalectomy (34), and in a 50:1 ratio in assays using the Shwartzman reaction and neutropenia in rabbits (8, 35) and hypotension in dogs (31). We used 100:1 and 200:1 ratios of LALF to LPS by weight in rabbits and sheep, respectively. Using these doses, there was a substantial decrease in the effect of LPS challenge in each assay. The neutralization in sheep was particularly pronounced in that LPS-induced pulmonary artery hypertension was almost eliminated and neutropenia was not seen. Our results are in agreement with the findings of Alpert et al., who recently reported that a

TEMPERATURE CHANGE

CHANGE IN LEUKOCYTE COUNT 20001

1.5i 1.0

Wi~~~~~~~~~~~~~0

IiI

0bS

0.5 b-2 000

10

0.0 -4000

-0.5-60

0

60

120

1860

-60

2 40

0

60

120

180

240

fime hIn)

tUme (Wn)

PULMONARY VASCULAR RESISTANCE

PULMONARY ARTERY PRESSURE

Si

50

4

40c

I